1

Audit Committees and Earnings Quality

Peter Baxter – University of the Sunshine Coast

Julie Cotter* – University of Southern Queensland *Address for correspondence - School of Accounting, Economics and Finance, Faculty of Business,

University of Southern Queensland, Toowoomba QLD 4350, AUSTRALIA

Abstract

This research investigates whether audit committees are associated with improved

earnings quality for a sample of Australian listed companies prior to the introduction of

mandatory audit committee requirements in 2003. Two measures of earnings quality are

used based on models first developed by Jones (1991) and Dechow and Dichev (2002).

Our results indicate that formation of an audit committee reduces intentional earnings

management but not accrual estimation errors. We also find differences in the

associations between audit committee accounting expertise and the two earnings quality

measures. Other audit committee characteristics examined are not significantly related to

either earnings quality measure.

Keywords: Audit committees, Corporate governance, Earnings management, Earnings

quality

JEL Descriptors: G30, G38, M41

_____________

This paper is from Peter‟s PhD thesis completed at the University of Southern Queensland (USQ). We are

grateful for the input of his Associate Supervisor, Gary Monroe from the Australian National University

(ANU). We also thank the two anonymous reviewers for their very helpful comments and suggestions.

This paper has also benefited from the comments of seminar participants at Griffith University, ANU, USQ,

Central Queensland University (CQU), the 2005 University of Technology Sydney Accounting Research

Summer School, and the 2005 Accounting and Finance Association of Australia and New Zealand

(AFAANZ) Conference. Funding support was provided by USQ, CQU and a PhD scholarship jointly

sponsored by AFAANZ, CPA Australia and the Institute of Chartered Accountants in Australia. The GICS

industry classification data was kindly provided by Standard and Poor‟s. The Global Industry Classification

Standard (“GICS”) was developed by and is the exclusive property and a trademark of Standard & Poor‟s, a

division of The McGraw-Hill Companies, Inc. (“S&P”) and Morgan Stanley Capital International Inc.

(“MSCI”).

2

1. Introduction

The purpose of this paper is to investigate the association between audit

committees and earnings quality in Australia. We examine two key aspects of this

relation, audit committee formation and audit committee characteristics. We use

measures of earnings quality based on models first developed by Jones (1991) and

Dechow and Dichev (2002). Measures based on the Jones „earnings management‟ model

are generally characterised as capturing managements‟ intent to manipulate earnings,

while measures based on Dechow and Dichev‟s „accrual estimation error‟ model include

accrual estimation errors arising from management lapses or environmental uncertainties.

Improved quality of financial reporting practices, and more specifically earnings,

has been widely cited as one of the major benefits of companies establishing audit

committees (Blue Ribbon Committee, 1999; Australian Accounting Research Foundation

(AARF) et al., 2001; Ramsay, 2001). However, the approach adopted by the Australian

Stock Exchange (ASX)1 from the early 1990s to 2003 was one of disclosure only,

requiring listed companies to provide statements about their main corporate governance

practices, including whether they had an audit committee and if appropriate, why they did

not comply with best practice guidelines. Audit committees only became mandatory in

2003 for those listed companies on the S & P All Ordinaries Index following the

recommendations of the ASX Corporate Governance Council2 (ASX Corporate

1 Following the merger of the Australian Stock Exchange with the Sydney Future Exchange in 2006, the

ASX became the Australian Securities Exchange. 2 A second edition of these recommendations was issued in 2007, but the 2003 edition applies to this study.

3

Governance Council, 2003).3 Given the previous relative lack of audit committee

regulation in Australia as compared to the US and other overseas jurisdictions4, pre-2003

Australia represents a rich empirical setting for the analysis of the association between

audit committees and earnings quality.

Davidson et al. (2005) and Koh et al. (2007) are the only known published studies

to utilise this voluntary institutional setting to explore the relationship between audit

committees and earnings quality. We extend their research in several ways. First, we

capture earnings quality using measures of accrual estimation errors as well as abnormal

accruals. The accrual estimation errors measure is a more comprehensive measure of

earnings quality. We are not aware of any prior published research into the relationship

between audit committees and earnings quality that uses measures based on Dechow and

Dichev‟s (2002) accrual estimation errors model. A comparison of our results between

these two earnings quality measures allows us to investigate the potential impact of audit

committees on different aspects of earnings quality.5 Second, we examine whether

earnings quality increases following the voluntary formation of an audit committee.

While several studies including Davidson et al. (2005) have examined whether the

existence of an audit committee is associated with earnings quality, tests of this

association do not differentiate between whether (a) the audit committee impacts earnings

3 In addition, entities in the top 300 of the Index are now required to comply with the ASX Corporate

Governance Council‟s best practice recommendations relating to the composition, operation and

responsibility of the audit committee (Australian Stock Exchange, 2006). 4 Audit committees have been mandatory on the major US stock exchanges since as early as 1978

(Vanasco, 1994). More recently, there has been an increasing trend around the world towards requiring

listed companies to not only establish audit committees, but also to ensure that they meet pre-specified

requirements including composition and reporting obligations. For example, in the US following

recommendations of the Blue Ribbon Committee (1999), the New York Stock Exchange and the National

Association of Securities Dealers changed their listing rules to require listed companies to maintain audit

committees with at least three directors, all of whom are independent of management (Klein, 2003). 5 Unpublished research by Dhaliwal et al. (2006) and Kent et al. (2008) use measures based on the Dechow

and Dichev (2002) model to capture accruals quality. However neither of these studies makes comparisons

between measures of accruals quality and earnings management.

4

quality or (b) firms with high quality earnings are more likely to form an audit

committee. Overseas research (Wild, 1994; Jeon et al., 2004) has found mixed evidence

about the impact of audit committee formation on earnings quality. Third, in addition to

the audit committee characteristics examined by Davidson et al. (2005) and Koh et al.

(2007), we investigate the impact of audit committee expertise on earnings quality.

Recent unpublished work in the US by Dhaliwal et al. (2006) reports an association

between audit committee accounting expertise and accruals quality. Finally, we use a

more refined measure of audit committee independence than that used in prior Australian

studies that investigate the association between audit committee characteristics and

earnings quality (Davidson et al., 2005; Koh et al., 2007).

Our results suggest that earnings quality increases in the year following voluntary

audit committee formation. However this is only the case when earnings quality is

captured using measures based on Jones‟ (1991) earning management model rather than

Dechow and Dichev‟s (2002) accrual estimation error model. This result appears to

indicate that audit committees are effective in reducing intentional accrual manipulations,

which are better captured by the Jones model. We also find differences in the

associations between audit committee accounting expertise and the two earnings quality

measures. When we capture earnings quality using accrual estimation errors, we find

higher earnings quality (lower accrual estimation errors) for companies with a greater

proportion of qualified accountants on their audit committee. However, we do not find a

similar reduction in earnings management. Indeed, we find some evidence that suggests

higher abnormal accruals for firms with a greater proportion of accounting expertise on

their audit committee. Results pertaining to our other audit committee characteristics are

5

similar to those found by Davidson et al. (2005) with the exception of audit committee

independence. Using our more refined measure of independence, we find that this audit

committee characteristic does not impact earnings quality.

The remainder of this paper is organised as follows: Section 2 outlines the prior

literature and hypotheses tested in this paper. Section 3 delineates our earnings quality

measures, while Section 4 describes the empirical analysis. Section 5 concludes the

paper.

2. Prior literature and hypotheses

2.1 Audit committee formation

Several prior studies provide empirical support for a cross-sectional association

between audit committees and financial reporting quality (e.g., McMullen, 1996; Dechow

et al., 1996; Beasley et al., 2000). However, the research designs used in these prior

studies are unable to establish whether the existence of an audit committee per se impacts

earnings quality. For a more direct test of the impact of audit committees on earnings

quality, it is necessary to consider changes in earnings quality subsequent to the

formation of an audit committee.

The only known published study that directly examines the association between

the formation of audit committees, earnings management and, inversely, earnings quality

is Jeon et al. (2004). Contrary to expectations, their findings indicate that earnings

management did not significantly decrease in the period after audit committee formation.

These results conflict with those of Wild (1994) who finds a significant increase in the

market's reaction to earnings reports released after audit committee formation.

6

We propose an association between the formation of an audit committee and an

increase in earnings quality. Tests will allow a direct assessment of whether the voluntary

formation of an audit committee is followed by an increase in earnings quality for our

sample of Australian companies.

H1: The formation of an audit committee is associated with an increase in

earnings quality.

2.2 Audit committee characteristics

Independence

The independence of an audit committee is often considered an essential

characteristic influencing the committee‟s effectiveness in overseeing the financial

reporting process. It can be argued that independent directors are in the best position to

serve as active overseers of the financial reporting process, thereby having a greater

ability to withstand pressure from management to manipulate earnings (Klein, 2002).

Audit committee independence has been found to be significantly associated with

measures of earnings quality in several prior studies (e.g., Klein, 2002; Bedard et al.,

2004; Choi et al., 2004; Van der Zahn and Tower, 2004; Davidson et al., 2005; Vafeas,

2005). However, within these studies, there are some inconsistencies in the results. For

example, Klein (2002) finds no evidence of a significant association between an audit

committee comprised solely of independent directors and her measure of earnings

management. Whereas, Bedard et al. (2004) find that the same measure of audit

committee independence is negatively associated with the likelihood of aggressive

earnings management.

7

Expertise

In addition to independence, the expertise of the audit committee is generally

considered an important characteristic for its effective operation. It has been argued that

effective oversight by an audit committee requires that its members possess sufficient

expertise in accounting and auditing to independently assess the matters that are

presented to them (Beasley and Salterio, 2001; Davidson et al., 2004; DeFond et al.,

2005).

Several prior studies have found a significant association between the expertise of

the audit committee and earnings quality (e.g., Xie et al., 2003; Bedard et al., 2004; Choi

et al., 2004; Dhaliwal et al., 2006). However, some inconsistencies exist between the

results of these studies and others such as Van der Zahn and Tower (2004) who failed to

find an association between the magnitude of earnings management and the audit

committee's financial expertise amongst the independent directors.

Activity and size

The level of activity of an audit committee has been recommended as important to

enhance its effectiveness in improving earnings quality. Menon and Williams (1994)

suggest that the mere formation of an audit committee does not mean that the committee

is actually relied on by the board of directors to enhance its monitoring ability. Choi et al.

(2004, p.41) argue that an "…actively functioning audit committee is more likely to

detect earnings management than a dormant committee." In addition, the size of an audit

committee can have a positive impact on earnings quality. Larger audit committees can

be more effective as they are likely to include members with varied expertise to perform

more intense monitoring of financial reporting practices (Choi et al., 2004).

8

Inconsistent results in the prior studies also exist for the association between audit

committee activity and earnings management or earnings quality. While Xie et al. (2003),

Van der Zahn and Tower (2004) and Vafeas (2005) find evidence of a significant

association between these variables, Choi et al. (2004), Bedard et al. (2004) and

Davidson et al. (2005) find that audit committee activity is not significantly related to

earnings management. Similar inconsistent results also exist in relation to the size of the

audit committee. We use the following hypothesis:

H2: The independence, expertise, activity, and size of an audit committee are

positively associated with earnings quality.

3. Earnings quality measures

3.1 Earnings quality vs earnings management

This paper uses two measures of earnings quality. The first measure uses a

modified version of the Jones (1991) model of discretionary accruals. This measure has

been widely used in the literature to capture earnings management, which can be viewed

as an inverse measure of earnings quality. Schipper (1989, p. 92) defines earnings

management as "…a purposeful intervention in the external financial reporting process,

with the intent of obtaining some private gain." Under this perspective, opportunistic

earnings management negatively impacts on the quality of earnings, i.e., the greater the

earnings management, the lower the earnings quality.6

Our second measure of earnings quality uses a modified version of the Dechow

and Dichev (2002) accrual estimation errors model. This model is based on the argument

that estimation errors in accruals and subsequent corrections of these errors decrease the

6 An alternative view is that earnings are managed to allow managers to reveal more private information to

users about the financial reports (Schipper, 1989; Healey and Wahlen, 1999).

9

quality of accruals and earnings. However, unlike the Jones (1991) type models of

discretionary accruals, no attempt is made to separate the intentional from the

unintentional accrual estimation errors (Dechow and Dichev, 2002). This is because both

types of errors imply low quality earnings.

3.2 Measures of Earnings Quality

We capture earnings quality using absolute value measures from the two models

described below. The sign of these measures is deemed not to be relevant since all

deviations from underlying earnings reduce earnings quality, regardless of their direction.

They are inverse measures of earnings quality. We use cross-sectional rather than time-

series specifications for each of our measures since we require measures of earnings

quality for specific firm years. Information on the Global Industry Classification Standard

(GICS) is used to form the industry matched samples required to calculate our earnings

quality variables. To ensure sufficient degrees of freedom and enhance the validity of

these measures, we limit our sample to companies in those industry groups that had 20 or

more companies listed on the ASX. For companies in large industry groups, our industry

matched samples comprise 30 companies.

Our first measure of earnings quality (EQJones) is based on the modified version

of the Jones (1991) discretionary accruals model proposed by Dechow et al. (1995).7 We

use cross-sectional samples of companies in the same industry groups as the sample

companies. The absolute value of discretionary accruals is used as our first measure of

earnings quality (EQJones).

7 This version of the Jones (1991) model includes the change in receivables in the equation used to estimate

the industry specific coefficients. Since this model is well established in the literature, we do not provide

further details about how we calculate discretionary accruals here.

10

It has been argued that there is the potential for discretionary accruals models to

misclassify expected accruals as unexpected because of the incompleteness of the

expected accruals model (Bernard and Skinner, 1996; Larcker and Richardson, 2004).

Guay et al. (1996) suggests that their evidence was consistent with the models estimating

discretionary accruals with considerable imprecision and/or misspecification. Hansen

(1999) concludes that studies relying entirely on the validity of discretionary accruals

models were likely to under- or overstate proposed earnings management behaviour.

Dechow et al. (1995) demonstrates that discretionary accruals models typically generated

tests of low power for earnings management of economically plausible magnitudes.

In an attempt to overcome criticisms of the modified Jones model, we use an

additional proxy for earnings quality. Our second measure of earnings quality (EQDD)

uses the cross-sectional version of the Dechow and Dichev (2002) accrual estimation

error model employed by Francis et al. (2005).8 McNichols (2002) provides a critique of

the Dechow and Dichev (DD) model9. Following McNichols‟ (2002) critique and

associated recommendations for improvement, Francis et al. (2005) add two variables

from the Jones (1991) model, i.e., the change in current sales and the level of property

plant and equipment.

We calculate EQDD by estimating the modified following regression for each

sample company relative to its industry group of companies for each of the years of

interest. All variables in equation (4) are divided by average total assets:

WCt = b0 + b1CFOt-1 + b2CFOt + b3CFOt+1 + b4Salest + b5PPEt +t (4)

8 Our results are essentially unchanged when the original Dechow and Dichev (2002) model is used.

9 McNichols (2002) identifies several specific areas of weakness with the DD model. These include a

failure to separately consider how total accruals might be affected by the behaviour of discretionary

accruals.

11

Where:

WCt = Working capital in year t i.e. Accounts receivable + Inventory -

Accounts payable - Taxes payable + Other assets (net);

CFOt-1 = Cash flows from operations in year t – 1;

CFOt = Cash flows from operations in year t;

CFOt+1 = Cash flows from operations year in year t + 1;

Salest = Sales in year t less sales in year t – 1;

PPEt = Gross property, plant and equipment in year t

This measure of earnings quality captures the extent to which accruals map into

cash flow realisations in past, present and future cash flows. Francis et al. (2005) use the

standard deviation of the residuals from this model as a measure of earnings quality.

However, we are not able to use the standard deviation of the residuals from our cross-

sectional industry model since this would provide a measure of earnings quality across all

companies in the industry group rather than just the company of interest. Following

Srinidhi and Gul (2007) who also need to capture this measure on a firm-year basis, we

use the absolute value of the residual as our measure of earnings quality. The higher the

absolute residual for each sample company, the lower is the quality of earnings.

4. Empirical analysis

4.1 Data and sample

The financial statement data items used to estimate our earnings quality measures

are extracted from the Aspect Financial Database (SIRCA Ltd, 2004). To facilitate testing

of hypothesis 1 which proposes an association between audit committee formation and an

12

increase in earnings quality, these variables are estimated for the years before and after

audit committee formation. That is, we use industry matched samples to estimate our

earnings quality measures for both the pre and post formation years. In addition, they are

re-estimated for each of our sample firms in 2001, since this is the year used to test the

associations between earnings quality and audit committee independence, expertise,

activity, and size proposed in hypothesis 2.10

Data required for these audit committee variables is hand collected from the 2001

annual reports. Audit committee independence and expertise for each director is assessed

from disclosures about directors‟ backgrounds, qualifications and experience. The

definition of director independence as specified by the ASX Corporate Governance

Council (2003) was used11

. Accounting and legal expertise are defined in terms of

professional qualifications.

[Insert table 1 here]

The sample is drawn from the top 500 Australian companies listed on the

Australian Stock Exchange (ASX) with financial years ending during 2001. Sample

selection procedures and final sample sizes for hypotheses tests are shown in Table 1.

We exclude companies without an audit committee (37) and those companies for which it

could not be determined whether an audit committee existed (4). Banks, trusts and

foreign companies (37) are also excluded since financial reporting requirements for these

companies differ from those of other companies listed on the ASX. Companies in the

10

This year is selected as the base year to avoid any effects of companies anticipating the new ASX listing

rule requiring audit committees to be formed by all companies in the S&P All Ordinaries Index. This new

rule came into effect from 1 January 2003. 11

Essentially, independent directors are non-executive directors who do not have a business or other

relationship with the firm that could interfere with their ability to act independently. These assessments

were made by one author based on annual report information and validated by the other.

13

Diversified Financials and Real Estate industry groups (15) are excluded because they do

not typically generate any sales revenue, which is needed to calculate our earnings quality

variables. As we require sufficiently large numbers of companies to form the industry

matched samples needed to calculate our measures of earning quality, we delete 74

companies from several small GICS industry groups12

. Finally, we delete 24 companies

where complete annual report data for 2001 is not available. This leaves a final sample

size of 309 companies for tests of the association between audit committee characteristics

and earnings quality (H2). Table 2 Panel A shows the industry breakdown of our sample.

[Insert table 2 here]

Further deletions from our sample are needed for tests of the association between

the formation of an audit committee and earnings quality (H1). In particular, we exclude

companies for which we are unable to reliably determine the audit committee formation

year from annual reports. These comprise companies whose audit committees were

formed prior to 1993 requirements to disclose audit committees in annual reports (80),

those that listed on the ASX with an audit committee already in place (133), and those for

which pre/post formation year annual report data is not available (24). This left a sample

of 72 companies for tests of hypothesis one. Panel B of Table 2 shows the number of

companies forming their audit committee by year. The higher numbers of formations

during the 1994 to 1996 period suggest that the 1993 introduction of disclosure

requirements provided an impetus for some companies to form an audit committee.

12

These industry groups were Automobiles and Components; Consumer Durables and Apparel; Food and

Staples Retailing; Household and Personal Products; Transportation; Insurance; Semiconductors and

Semiconductor Equipment; and Utilities.

14

4.2 Audit committee formation and earnings quality

To determine the effect of audit committee formation on earnings quality, we

compare our earnings quality measures between the years before and after each

company‟s audit committee was formed. Panel A of Table 3 shows the results of

matched-pairs t-tests for significant differences for these accruals measures pre and post

audit committee formation.13

For the accruals levels variables derived from the modified

Jones (1991) model, the mean for EQJones(post) (0.1370) is significantly less than the

mean for EQJones(pre) (0.2033). This result suggests that earnings quality calculated

based on the Jones (1991) model is significantly higher in the year after formation of the

audit committees compared to the year before audit committee formation. These results

support our first hypothesis that the formation of an audit committee is associated with an

increase in earnings quality.

[Insert table 3 here]

However, the results for the measure of earnings quality based on the Dechow and

Dichev (2002) model do not show a significant difference between the years before and

after audit committee formation. Correlation coefficients between EQJones and EQDD

are not significant (see Table 5), indicating that these two measures capture quite

different aspects of earnings quality. It is possible that the observed change in EQJones

between the pre and post formation years is due to factors other than the formation of the

audit committee, such as changes in the board and auditor. To control for the impact of

13

A preliminary analysis of the distributions for our earnings quality variables revealed a small number of

extreme outliers as well as positive skewness. Three extreme outliers are excluded from the analysis for

EQJones, while one is excluded for EQDD. Wilcoxon signed ranks tests using the full sample yield the

same inferences, as do sensitivity tests using logged transformations.

15

these potentially correlated omitted variables on the relationship between earnings quality

and audit committee formation, the following pooled regression is estimated:

EQ = a + b0 FORMATION + b1 ROA + b2 BDIND + b3 BDACCEX + b4 BDLEGEX

+ b 5 BDCMEET + b6 BDSIZE + b7 AUDITOR + (5)

FORMATION is a dummy variable that equals zero in the pre formation year

and one in the post formation year. Each of our control variables is measured in both the

pre and post formation years. Return on assets (ROA) is included to control for potential

changes in firm performance. It is possible that the observed increase in earnings quality

could be associated with a change in firm performance. Prior research has shown that the

measurement of discretionary accruals can be problematic for firms with extreme

financial performance (Dechow et al., 1995; Kothari et al., 2005). It is also possible that

changes to the board of directors or company auditor occurring at the same time that the

audit committees were formed could be associated with the increase in earnings quality.

Hence, we include controls for board independence (BDIND), size (BDSIZE), accounting

expertise (BDACCEX), legal expertise (BDLEGEX), meetings per year (BDMEET), and

auditor quality (AUDITOR) for both the pre and post audit committee formation years.

Results of these pooled regressions are shown in Panel B of Table 3. The results

indicate that audit committee formation remains significantly associated with EQJones

when these other potential explanations are controlled. The negative coefficient on

FORMATION indicates that when this variable equals one (the post audit committee

formation year), EQJones is lower; thus indicating less earnings management and hence

16

higher earnings quality. ROA and BDMEET are also significantly negatively associated

with EQJones. None of these variables are significantly correlated with EQDD.14

Our EQJones results support those of Wild (1994) who finds a significant increase

in the market reaction to earnings reports released after the formation of the audit

committee. However they are inconsistent with the results of Jeon et al. (2004) who find

no significant decrease in earnings management for Korean firms after they established

audit committees. A potential reason for the inconsistency between our results and those

of Jeon et al. is the different legal environments between Korea and Australia. Their

sample included a majority that were required by Korean government law to establish an

audit committee. The period of study for our paper was prior to the mandatory

requirement for audit committee formation by large Australian listed companies, which

came into effect on 1 January 2003. Companies that form audit committees voluntarily,

not because of a government requirement, are likely to be more effective at constraining

earnings management and therefore improving earnings quality. This is because they

have other incentives to ensure their audit committees operate effectively, which also

drive the decision to voluntarily form an audit committee.

4.3 Audit committee characteristics and earnings quality

Table 4 provides the descriptive statistics for the variables used in the tests of

association between audit committee characteristics and earnings quality (H2) as well as

several control variables relevant to this association. The mean and median values for

EQJones are similar to those reported by Davidson et al. (2005) for their absolute

14

Extreme outliers are excluded for these tests. Results of sensitivity tests using logged transformations of

our EQ variables yield the same inferences about the significance relationship between audit committee

formation and EQJones.

17

discretionary accruals measure that is based on the same cross-sectional modified Jones

model that we use. We exclude several outliers for EQJones and EQDD from our primary

analysis and also report results of sensitivity analysis using logged transformations of our

earnings quality measures.

Overall, the descriptive statistics indicate that there is considerable variation in the

audit committee variables for the sample companies. The mean proportion of independent

directors on the audit committee is 0.53. Prior US studies such as Yang and Krishnan

(2005) provide evidence that audit committees in the United States have much higher

proportions of independent directors, which reflects the greater degree of audit committee

regulation. Our measures of ACMEET and ACSIZE are slightly higher than those

reported by Davidson et al. (2005). This is most likely due to the larger average size of

the firms in our sample and the exclusion of firms without an audit committee from our

sample. Descriptive statistics for full board level variables that correspond to our audit

committee variables are also shown in Table 4. Davidson et al. (2005) and Koh et al.

(2007) found board independence to impact earnings quality. It is likely that some of the

other board level variables are also associated with earnings quality. The remaining

variables in Table 4 are controls for auditor quality, leverage, firm size, losses and

operating cycle.

[Insert table 4 here]

Dechow and Dichev (2002) identify several innate factors that affect accruals

quality: firm size, the incidence of losses, operating cycle, and volatility of operating cash

flows and sales. Our sample includes firms ranging in size from total assets of $3.94M to

$84.96B, with a mean of $1.28B. The distribution of total assets is highly positively

18

skewed and we therefore take a log transformation of this variable (LNTA). LOSS

equals 1 if income for the year is less than zero, 0 otherwise. 108 of the sample firms

report a loss in 2001. Length of operating cycle is measured as 360/(sales/average

account receivables). Operating cycles for our sample firms range between 0 and 1050

days, with a mean of 65.68 days. This variable is highly positively skewed and we

therefore use a log transformation for our hypotheses tests (LNOPCYCLE). We do not

include controls for volatility of operating cash flows or sales since we are unable to

obtain a sufficient time-series of data to calculate these measures for the majority of our

sample firms.

Table 5 shows Pearson and Spearman correlation coefficients between the

earnings quality, audit committee, full board and control variables. For EQJones,

Pearson correlations show significant positive relationships with LOSS and LNOC, while

Spearman correlations show significant relationships with ACACCEX (+), BDIND (-),

BDACCEX (+), BDSIZE (-) and LOSS (+). The Spearman correlations between

EQJones and both ACACCEX and BDACCEX are positive rather than negative as

expected. This result appears to suggest that accounting expertise could be related to an

increase rather than a decrease in earnings management. When we use a log

transformation of EQJones, Pearson correlations with ACACCEX, BDACCEX, LOSS

and LNOC are all positive and significant, while BDSIZE is significantly negatively

associated with EQJones. Overall, these results do not support the relations between

EQJones and the audit committee characteristics predicted in H2.

[Insert table 5 here]

19

When we consider EQDD, Pearson correlations show significant negative

relationships between this measure of earnings quality and ACACCEX, ACSIZE,

BDSIZE, LNTA, and a significant positive relation with LOSS. Spearman correlations

support these results and also show a significant positive relation between EQDD and

LNOC. When we use a log transformation of EQDD, the same variables remain

significant. These results indicate initial support for the predicted H2 relations between

earnings quality and audit committee size and accounting expertise.

Not surprisingly, most of our audit committee and full board level variables are

very highly correlated; with the correlation coefficients for the independence and

expertise measures ranging between 0.69 and 0.78. Further, audit committee size is

significantly positively correlated with full board size and firm size. Interestingly, the

two measures of audit committee expertise (ACACCEX and ACLEGEX) are

significantly negatively correlated with each other. This suggests that the two forms of

expertise are substitutes for each other.

We use the following regression model to test our second hypothesis that earnings

quality is positively associated with audit committee independence, expertise, activity

and size. EQ denotes the two earnings quality measures described above (EQJones and

EQDD). This model is estimated on our sample of listed Australian companies in 2001:

EQ = a + b1 ACIND + b2 ACACCEX + b3 ACLEGEX + b4 ACMEET +

b5 ACSIZE + b6 AUDITOR + b7 LNTA + b8 LEV + b9 LOSS + b10LNOC +

(6)

In addition, we run the above model substituting a series of industry dummy

variables for LNOC. This allows us to use a larger sample since we were able to collect

20

data about industry membership for all of our sample firms, while we were only able to

obtain operating cycle data for 284 of our sample firms. We rerun this model controlling

for full board independence, expertise, activity and size. Several of these variables are

significantly positively correlated with their corresponding audit committee measures and

that is why we exclude them from equation 6. However, some of these board variables

are significantly associated with our EQ measures and we therefore attempt to control for

their impact by including them in a sensitivity test of this model.

Table 6 shows the results from OLS regressions of equation 6. None of our audit

committee variables are significantly associated with EQJones. Similarly, Davidson et al.

(2005) report insignificant coefficients for ACMEET and ACSIZE, and mixed results for

ACIND depending on how it is measured.15

Our results indicate that EQDD is

significantly negatively correlated with ACACCEX indicating that this measure of

earnings quality is higher when there are a greater proportion of audit committee

members with accounting expertise. This result is consistent with Dhaliwal et al. (2006)

who find a significant positive relation between accounting expertise and accruals

quality. Our other audit committee variables are not significantly related to EQDD.16

[Insert table 6 here]

When logged transformations of our EQ variables are used, EQDD remains

significantly negatively associated with ACACCEX, while EQJones is significantly

15

These authors proxy audit committee independence using a dichotomous non-executive director measure

and find mixed results depending on whether they code this variable with a value of one if the audit

committee is comprised entirely of non-executive directors or a majority. In sensitivity tests, their

significant results for this variable become insignificant when they remove non-executive directors that had

related party transactions. 16

We also examine a summary measure of the overall strength of the sample companies' audit committees.

This variable (AC_GOV_SCORE) is calculated as the sum of each of the audit committee dichotomous

variables discussed above. There is a significant negative Pearson correlation between AC_GOV_SCORE

and EQDD. However, this relation is not significant in a multivariate context.

21

positively associated with this variable. When we add the full board variables to our

models, the relationship between EQDD and ACACCEX becomes insignificant and the

remainder of our results are qualitatively the same. Given the high correlation between

our board and audit committee accounting expertise variables (r = 0.77), it is difficult to

reliably interpret this result. We therefore rerun our EQDD models with BDACCEX

instead of ACACCEX and find that BDACCEX is not significantly related to EQDD.

This result suggests that it is accounting expertise at the audit committee level rather than

the full board level that positively impacts earnings quality.

The results for control variables shown in table 6 indicate significant associations

between EQDD and LNTA, and between EQJones and LNTA and AUDITOR, as well as

some mixed results for LEV, LOSS and LNOC. The significant positive relations that we

observe between EQJones and ACACCEX and AUDITOR are contrary to expectations.

Several of the industry dummy variables are significant for EQJones, which captures

variation in the exercise of discretionary accruals across industries.

Overall, H2 is generally not supported, with the exception of audit committee

accounting expertise when the EQDD measure of earnings quality is considered. The

weight of evidence suggests that the higher the proportion of accounting expertise a

company has on its audit committee, the lower its accrual estimation errors.

5. Conclusions

This research investigates the association between audit committees and earnings

quality in Australia. The time period for the research is selected to avoid the confounding

effects of mandatory audit committee requirements introduced for Australian companies

in 2003. We hypothesise that the formation of an audit committee is associated with an

22

increase in earnings quality (H1); and the independence, expertise, activity, and size of an

audit committee are positively associated with earnings quality (H2). Overall, the results

provide support for H1, but not H2.

Several conclusions can be drawn from our results. First, we find that a

discretionary accruals measure based on the Jones (1991) earnings management model,

decreases significantly in the year following audit committee formation. Since measures

based on this model are generally characterised as capturing managements‟ intent to

manipulate earnings, our results imply that the establishment of an audit committee is an

effective way to reduce earnings management, and hence improve the quality of earnings.

When we capture accrual estimation errors using measures based on Dechow and

Dichev‟s (2002) model, we do not find an increase in earnings quality following audit

committee formation. This disparity in results between the two types of earnings quality

measures highlights the potential impact of audit committees. While improved quality of

financial reporting practices has been widely cited as a major benefit of audit committees,

this result appears to indicate that this improvement most likely occurs through a

reduction in earnings manipulations rather than lower accrual estimation errors deriving

from management lapses or environmental uncertainties. A caveat on these results is the

relatively small sample size available for tests of H1.

Second, when we capture earnings quality using an accrual estimation errors

measure, we find that audit committee accounting expertise is associated with higher

quality earnings. However we do not find the same association when we capture earnings

quality using an earnings management measure. Indeed, we find some evidence of higher

earnings management for firms with a greater proportion of qualified accountants on their

23

audit committees. Future research that explores this result further may be able to shed

some light on this unexpected finding. A potential limitation of our research relates to the

endogeneity of audit committees. The characteristics of audit committees are not

necessarily independent of earnings quality. Companies with higher quality earnings

may be more likely to choose audit committee characteristics that signal the strength of

their financial reporting system (Engel, 2005).

Overall, our results highlight the multifaceted nature of earnings quality and the

potential for audit committees to impact it. As we have found, different measures of

earnings quality can lead to different results and inferences. Each of the available models

of earnings quality has its own particular limitations and these should be considered when

interpreting our results. Additional research that separates out the intentional and

unintentional components of the accrual estimation errors would help to further clarify

which aspects of earnings quality audit committees tend to improve.

24

References

Australian Accounting Research Foundation, Institute of Internal Auditors Australia and

Australian Institute of Company Directors, 2001, Best Practice Guide: Audit

Committees.

ASX Corporate Governance Council, 2003, Principles of Good Corporate Governance

and Best Practice Recommendations.

Australian Stock Exchange, 2006, ASX listing rules (Retrieved 4 August 2006 from

http://www.asx.com.au/supervision/rules_guidance/listing _rules1.htm).

Beasley, M., J. Carcello, D. Hermanson and P. Lapides, 2000, Fraudulent financial

reporting: consideration of industry traits and corporate governance mechanisms,

Accounting Horizons 14, 441-454.

Beasley, M. and S. Salterio, 2001, The relationship between board characteristics and

voluntary improvements in audit committee composition and experience,

Contemporary Accounting Research 18, 529-570.

Bedard, J., S. Chtourou and L. Courteau, 2004, The effect of audit committee expertise,

independence and activity on aggressive earnings management, Auditing: A Journal

of Practice and Theory 23, 13-35.

Bernard, V. and D. Skinner, 1996, What motivates managers‟ choice of discretionary

accruals?, Journal of Accounting and Economics 22, 313-325.

Blue Ribbon Committee, 1999, Report and recommendations of the Blue Ribbon

Committee on Improving the Effectiveness of Corporate Audit Committees, The

Business Lawyer 54, 1067-1095.

Choi, J., K. Jeon and J. Park, 2004, The role of audit committees in decreasing earnings

management: Korean evidence, International Journal of Accounting, Auditing and

Performance Evaluation 1, 37-60.

Davidson, R., J. Goodwin-Stewart and P. Kent, 2005, Internal governance structures and

earnings management, Accounting and Finance 45, 241-268.

Davidson, W., B. Xie and W. Xu, 2004, Market reaction to voluntary announcements of

audit committee appointments: the effect of financial expertise, Journal of Accounting

and Public Policy 23, 279-293.

Dechow, P. and I. Dichev, 2002, The quality of accruals and earnings: the role of accrual

estimation errors, The Accounting Review 77, 35-59.

Dechow, P., R. Sloan and A. Sweeney, 1995, Detecting earnings management, The

Accounting Review 70, 193-225.

Dechow, P., R. Sloan and A. Sweeney, 1996, Causes and consequences of earnings

manipulations: an analysis of firms subject to enforcement actions by the SEC,

Contemporary Accounting Research 13, 1-36.

DeFond, M., R. Hann and X. Xu, 2005, Does the market value financial expertise on

audit committees of boards of directors?, Journal of Accounting Research 53, 153-

193.

Dhaliwal, D., V. Naiker and F. Navissi, 2006, Audit committee financial expertise,

corporate governance and accruals quality: an empirical analysis, Working paper

(University of Arizona, University of Auckland and Monash University).

Engel, E., 2005, Discussion of „Does the market value financial expertise on audit

committees of boards of directors?‟, Journal of Accounting Research 43, 195-204.

25

Francis, J., R. LaFond, P. Olsson and K. Schipper, 2005, The market pricing of accruals

quality, Journal of Accounting and Economics 39, 295-337.

Guay, W., S. Kothari and R. Watts, 1996, A market-based evaluation of discretionary

accrual models, Journal of Accounting Research 34, 83-105.

Hansen, G., 1999, Bias and measurement error in discretionary accrual models, Working

paper (Penn State University).

Healy, P. and J. Wahlen, 1999, A review of earnings management literature and its

implications for standard setting, Accounting Horizons 13, 365-383.

Jeon, K., J. Choi and J. Park, 2004, The relationship between new audit committee

establishment and earnings management, Korean Accounting Review 29, 143-178.

Jones, J., 1991, Earnings management during import relief investigations, Journal of

Accounting Research 29, 193-228.

Kalbers, L. and T. Fogarty, 1998, Organisational and economic explanations for audit

committee oversight, Journal of Managerial Issues 10, 129-150.

Kent, P., J. Routledge and J. Stewart, 2008, Innate and discretionary accrual quality and

corporate governance, Paper presented at the AFAANZ Annual Conference, Sydney.

Klein, A., 2002, Audit committee, board of director characteristics and earnings

management, Journal of Accounting and Economics 33, 129-150.

Klein, A., 2003, Likely effects of stock exchange governance proposals and Sarbanes-

Oxley on corporate boards and financial reporting, Accounting Horizons 17, 343-355.

Koh, P., S. Laplante and Y. Tong, 2007, Accountability and value enhancement roles of

corporate governance, Accounting and Finance 47, 305-333.

Kothari, S., A. Leone and C. Wasley, 2005, Performance matched discretionary accrual

measures, Journal of Accounting and Economics 39, 163-197.

Larcker, D. and S. Richardson, 2004, Fees paid to audit firms, accrual choices and

corporate governance, Journal of Accounting Research 42, 625-658.

McMullen, D., 1996, Audit committee performance: an investigation of the consequences

associated with audit committees, Auditing: A Journal of Practice and Theory 15, 87-

103.

McNichols, M., 2002, Discussion of „The quality of accruals and earnings: the role of

accrual estimation errors‟, The Accounting Review 77, 61-69.

Menon, K. and D. Williams, 1994, The use of audit committee for monitoring, Journal of

Accounting and Public Policy 13, 121-139.

Rahman, R. and F. Ali, 2006, Board, audit committee, culture and earnings management:

Malaysian evidence, Managerial Auditing Journal 21, 783-804.

Ramsay, I., 2001, Independence of Australian Company Auditors: Review of Current

Australian Requirements and Proposals for Reform (Commonwealth of Australia,

Canberra, ACT).

Schipper, K., 1989, Commentary on earnings management, Accounting Horizons 36, 91-

102.

SIRCA Ltd, 2004, Aspect financial database (Retrieved 4 August 2006 from

http://www.sirca.org.au/uni/services/Data/accounting/aspect_financial.html)

Srinidhi, B. and F. Gul, 2007, The differential effects of auditors' non-audit and audit fees

on accrual quality, Contemporary Accounting Research 24, 595-629.

Tabachnick, B. and L. Fidell, 1989, Using Multivariate Statistics (Harper Collins, New

York).

26

Vafeas, N., 2005, Audit committees, boards and the quality of reported earnings,

Contemporary Accounting Research 22, 1093-1122.

Van der Zahn, M. and G. Tower, 2004, Audit committee features and earnings

management: further evidence from Singapore, International Journal of Business

Governance and Ethics 1, 233-257.

Vanasco, R., 1994, The audit committee: an international perspective, Managerial

Auditing Journal, 9, 18-42.

Vicknair, D., K. Hickman and K. Carnes, 1993, A note on audit committee independence:

evidence from the NYSE on „grey‟ area directors, Accounting Horizons 7, 53-57.

Wild, J., 1994, The audit committee and earnings quality, Journal of Accounting,

Auditing and Finance 22, 247-276.

Xie, B., W. Davidson and P. DaDalt, 2003, Earnings management and corporate

governance: the role of the board and audit committee, Journal of Corporate Finance

9, 295-316.

Yang, J. and J. Krishnan, 2005, Audit committees and quarterly earnings management,

International Journal of Auditing 9, 201-219.

27

Table 1

Summary of sample sizes used for hypotheses tests

Top 500 ASX listed companies in 2001 500

Less,

-Companies without audit committees 37

-Audit committee existence could not be determined 4

- Banks, trusts and foreign companies 37

- Diversified financials and real estate 15

- Companies from small four digit GICS industry groups 74 167

333

Less, Complete annual report data for 2001 not available 24

Sample for audit committee characteristics tests (H2) 309

Less,

-Audit committee formed prior to 1993

-Listed with audit committee in place

-Complete annual report data for pre/post audit

committee formation years not available

80

133

24

237

Sample for audit committee formation tests (H1) 72

28

Table 2

Panel A: Sample of 309 companies used for audit committee characteristics tests by

industry group

Industry group Number Percentage

Capital goods 33 10.7

Commercial services and supplies 21 6.8

Energy 20 6.5

Food, beverage and tobacco 29 9.4

Healthcare equipment and services 18 5.8

Hotels, restaurants and leisure 14 4.5

Materials 71 23.0

Media 20 6.5

Pharmaceuticals and biotechnology 16 5.2

Retailing 21 6.8

Software and services 25 8.1

Technology hardware and equipment 10 3.2

Telecommunication services 11 3.5

Total 309 100

Panel B: Number of audit committees formed each year by 72 ASX listed companies

that formed their audit committees following the 1993 requirements for audit

committee disclosures.

Year of audit committee formation Number of companies

1993 6

1994 14

1995 12

1996 15

1997 4

1998 9

1999 6

2000 6

Total 72

29

Table 3

Comparisons of earnings quality for the years pre and post audit committee

formation for 72 ASX listed companies

Panel A: Matched-pairs t-tests

Variable N Min. Max. Median Mean Std.

Dev.

t

EQJones(pre) 69 0.01 0.83 0.1209 0.2033 0.2046 3.058**

EQJones(post) 69 0.00 0.83 0.0923 0.1370 0.1444

EQDD(pre) 71 0.00 0.50 0.0561 0.0906 0.1047 -0.300

EQDD(post) 71 0.00 0.72 0.0580 0.0961 0.1199

Panel B: Pooled regression results Variable Pred. sign EQJones EQDD

Intercept 0.284

(4.187)**

0.098

(2.220)*

FORMATION

-

-0.066

(-2.254)*

0.007

(0.350)

ROA - -0.160

(-2.512)**

-0.047

(-1.121)

BDIND - 0.080

(1.276)

0.047

(1.123)

BDACCEX - -0.002

(-0.018)

0.025

(0.421)

BDLEGEX - 0.103

(0.986)

0.060

(0.881)

BDMEET - -0.006

(-2.230)*

-0.003

(-1.494)

BDSIZE - -0.003

(-0.390)

-0.002

(-0.358)

AUDITOR - -0.052

(-1.647)

0.007

(0.363)

Adjusted R2 0.098 -0.006

F statistic 2.867** 0.901

N 138 142

* significant at the 0.05 level, ** significant at the 0.01 level (p-values are one-tailed)

EQJones = Cross sectional earnings quality proxy from modified Jones (1991) model (i.e., absolute value

of abnormal accruals)

EQDD = Cross sectional earnings quality proxy from Dechow and Dichev (2002) adjusted for Jones (1991)

model variables (i.e., absolute value of regression residuals)

FORMATION: 1 = year after audit committee formation; 0 = year before audit committee formation

ROA = Return on assets calculated as operating profit after tax scaled by average total assets

BDIND = Proportion of independent directors on the board

BDSIZE = Number of board members

BDACCEX = Proportion of directors on the board with accounting qualifications

BDLEGEX = Proportion of directors on the board with legal qualifications

BDMEET = Number of board meetings per annum

AUDITOR: 1 = Big 5 or 6 auditor; 0 = Non-big 5 or 6 auditor

30

Table 4

Descriptive statistics for 309 Australian listed companies in 2001

Panel A Continuous variables

Variable Minimum Maximum Median Mean Std Dev Skewness

EQJones 0.00 2.66 0.09 0.18 0.25 4.32

EQDD 0.00 1.29 0.05 0.10 0.15 4.83

ACIND 0.00 1.00 0.50 0.53 0.34 -0.12

ACACCEX 0.00 1.00 0.33 0.31 0.30 0.74

ACLEGEX 0.00 1.00 0.00 0.13 0.20 1.49

ACMEET 0.00 13.00 3.00 3.06 1.60 1.74

ACSIZE 2.00 7.00 3.00 3.18 1.00 1.16

BDIND 0.00 1.00 0.40 0.42 0.25 0.04

BDACCEX 0.00 0.80 0.20 0.22 0.18 0.65

BDLEGEX 0.00 0.50 0.09 0.11 0.12 0.99

BDMEET 3.00 33.00 11.00 11.34 4.28 0.88

BDSIZE 3.00 17.00 6.00 6.33 2.23 1.55

TA ($M) 3.94 84,961.00 138.28 1,276.30 6,020.56 10.60

LNTA 15.19 25.17 18.74 19.01 1.77 0.55

LEV 0.00 2.52 0.47 0.46 0.26 2.34

OPCYCLE 1.00 1050.00 48.00 65.68 96.94 6.32

LNOC 0.00 6.96 3.87 3.72 1.02 -0.95

Panel B Dichotomous variables

Variable Frequency of 1s Frequency of 0s

AUDITOR 247 (79.9%) 62 (20.1%)

LOSS 108 (34.9%) 201 (65.1%)

EQJones = Cross sectional earnings quality proxy from modified Jones (1991) model (i.e., absolute value

of abnormal accruals)

EQDD = Cross sectional earnings quality proxy from Dechow and Dichev (2002) model adjusted for Jones

(1991) model variables (i.e., absolute value of regression residuals)

ACIND = Proportion of independent directors on audit committee

ACACCEX = Proportion of directors on audit committee with accounting qualifications

ACLEGEX = Proportion of directors on audit committee with legal qualifications

ACMEET = Number of audit committee meetings for the year

ACSIZE = Number of audit committee members

BDIND = Proportion of independent directors on the board

BDACCEX = Proportion of directors on the board with accounting qualifications

BDLEGEX = Proportion of directors on the board with legal qualifications

BDMEET = Number of board meetings for the year

BDSIZE = Number of board members

TA = Total assets

LNTA = Natural log of total assets

LEV = Total liabilities divided by total assets

OPCYCLE = Operating cycle measured as 360/(sales/average account receivables)

LNOC = Natural log of operating cycle, measured as 360/(sales/average account receivables)

AUDITOR: 1 = Big 5 or 6 auditor; 0 = Non-big 5 or 6 auditor

LOSS: 1 = net income for the year is less than zero; 0 otherwise

31

Table 5

Pearson and Spearman Correlations for 309 Australian listed companies in 2001 (Pearson correlations are above diagonal; p

values are shown in parenthesis)

EQJones EQDD ACInd ACAccEx ACLegEx ACMeet ACSize BDInd BDAccEx BDLegEx BDMeet BDSize LNTA Lev Auditor Loss LNOC

EQJones - 0.01

(0.818)

-0.05

(0.370)

0.06

(0.319)

-0.05

(0.343)

-0.07

(0.195)

-0.01

(0.226)

-0.08

(0.148)

0.09

(0.119)

-0.05

(0.369)

0.04

(0.537)

-0.10

(0.079)

-0.09

(0.114)

0.06

(0.286)

0.10

(0.071)

0.11*

(0.047)

0.13*

(0.035)

EQDD 0.07 (0.127)

- 0.03 (0.560)

-0.14* (0.015)

-0.04 (0.468)

-0.03 (0.567)

-0.11* (0.048)

0.05 (0.347)

-0.11 (0.053)

0.00 (0.998)

0.01 (0.802)

-0.15** (0.008)

-0.24** (0.000)

-0.03 (0.632)

-0.05 (0.394)

0.17** (0.003)

0.04 (0.522)

ACInd -0.07 (0.252)

-0.02 (0.772)

- -0.10 (0.09)

-0.07 (0.235)

0.19** (0.001)

0.03 (0.668)

0.77** (0.000)

-0.14* (0.018)

-0.18** (0.002)

0.07 (0.252)

0.16** (0.006)

0.18** (0.002)

-0.01 (0.899)

0.12* (0.039)

-0.09 (0.108)

-0.05 (0.423)

ACAccEx 0.13* (0.025)

-0.12* (0.043)

-0.13* (0.018)

- -0.20** (0.000)

-0.03 (0.554)

-0.10 (0.082)

-0.12* (0.042)

0.77** (0.000)

-0.13* (0.024)

0.05 (0.356)

-0.03 (0.619)

-0.00 (0.983)

0.08 (0.175)

-0.12* (0.029)

-0.02 (0.782)

-0.06 (0.290)

ACLegEx 0.01 (0.858)

0.01 (0.943)

-0.07 (0.213)

-0.18** (0.002)

- 0.04 (0.534)

-0.05 (0.391)

-0.12* (0.040)

-0.17** (0.003)

-0.69** (0.000)

-0.05 (0.396)

0.10 (0.075)

0.16** (0.004)

-0.02 (0.788)

0.04 (0.499)

-0.01 (0.867)

0.05 (0.414)

ACMeet -0.07 (0.242)

-0.04 (0.503)

0.20** (0.000)

-0.01 (0.824)

0.09 (0.117)

- 0.21** (0.000)

-0.04 (0.540)

-0.04 (0.540)

0.04 (0.477)

0.15** (0.008)

0.34** (0.000)

0.39** (0.000)

0.10 (0.083)

0.15** (0.009)

-0.20** (0.001)

-0.04 (0.284)

ACSize -0.08 (0.170)

-0.12* (0.043)

0.03 (0.547)

-0.07 (0.250)

0.04 (0.484)

0.16** (0.004)

- 0.03 (0.633)

0.03 (0.633)

-0.06 (0.294)

0.09 (0.110)

0.33** (0.000)

0.23** (0.000)

0.12* (0.042)

0.06 (0.315)

-0.13* (0.028)

0.11 (0.059)

BDInd -0.11* (0.045)

-0.02 (0.690)

0.77** (0.000)

-0.14* (0.018)

-0.10 (0.074)

0.22** (0.000)

0.12* (0.035)

- -0.16** (0.005)

-0.15** (0.009)

0.07 (0.223)

0.18** (0.002)

0.26** (0.000)

0.01 (0.928)

0.18** (0.001)

-0.10 (0.096)

-0.01 (0.888)

BDAccEx 0.13* (0.021)

-0.04 (0.498)

-0.16** (0.006)

0.78** (0.000)

-0.13 (0.020)

-0.01 (0.900)

0.07 (0.258)

-0.16** (0.006)

- -0.15** (0.008)

-0.12* (0.043)

-0.12* (0.043)

-0.04 (0.519)

0.06 (0.323)

-0.14* (0.015)

0.02 (0.775)

-0.07 (0.262)

BDLegEx 0.02 (0.724)

-0.00 (0.974)

-0.16** (0.005)

-0.12* (0.035)

0.70** (0.000)

0.09 (0.115)

-0.01 (0.860)

-0.13* (0.027)

-0.11* (0.045)

- -0.01 (0.883)

-0.01 (0.883)

0.09 (0.106)

0.09 (0.132)

-0.02 (0.676)

0.04 (0.528)

0.05 (0.428)

BDMeet 0.01 (0.905)

-0.01 (0.917)

0.10 (0.074)

0.07 (0.252)

-0.03 (0.617)

0.22** (0.000)

0.13* (0.018)

0.11 (0.063)

0.11 (0.051)

-0.05 (0.405)

- 0.01 (0.899)

0.09 (0.132)

0.11* (0.046)

-0.01 (0.819)

0.04 (0.535)

-0.01 (0.824)

BDSize -0.13* (0.027)

-0.19** (0.001)

0.15** (0.009)

-0.01 (0.861)

0.14* (0.012)

0.28** (0.000)

0.36** (0.000)

0.17** (0.002)

-0.09 (0.134)

0.05 (0.364)

0.03 (0.565)

- 0.26** (0.000)

0.12* (0.030)

0.21** (0.000)

-0.17** (0.003)

0.04 (0.284)

LNTA -0.09 (0.114)

-0.21** (0.000)

0.16** (0.005)

0.01 (0.852)

0.18** (0.002)

0.43** (0.000)

0.25** (0.000)

0.25** (0.000)

-0.00 (0.953)

0.12* (0.031)

0.12* (0.032)

0.52** (0.000)

- .40** (0.000)

0.34** (0.000)

-0.32** (0.000)

-0.04 (0.480)

Lev 0.03 (0.567)

-0.00 (0.972)

-0.03 (0.636)

0.07 (0.209)

0.06 (0.268)

0.11 (0.054)

0.15** (0.008)

-0.00 (0.950)

0.07 (0.229)

0.09 (0.123)

0.16** (0.005)

0.16* (0.005)

0.42** (0.000)

- 0.14* (0.016)

-0.18** (0.002)

-0.04 (0.494)

Auditor 0.09 (0.098)

-0.06 (0.314)

0.11* (0.047)

-0.11 (0.055)

0.06 (0.287)

0.16** (0.006)

0.10 (0.089)

0.17** (0.003)

-0.10 (0.091)

0.01 (0.898)

-0.01 (0.907)

0.23** (0.000)

0.34** (0.000)

0.14* (0.018)

- -0.09 (0.113)

-0.04 (0.475)

Loss 0.18** (0.002)

0.19** (0.001)

-0.09 (0.123)

-0.02 (0.703)

0.01 (0.901)

-0.22** (0.000)

-0.11* (0.046)

-0.09 (0.117)

-0.01 (0.901)

0.05 (0.353)

-0.01 (0.904)

-0.20** (0.000)

-0.34** (0.000)

-0.17** (0.002)

-0.09 (0.113)

- 0.21** (0.000)

LNOC 0.08 (0.173)

0.13* (0.025)

-0.04 (0.501)

0.01 (0.823)

0.01 (0.850)

-0.10 (0.109)

0.09 (0.140)

-0.03 (0.679)

0.06 (0.314)

0.01 (0.833)

-0.02 (0.748)

-0.02 (0.780)

-0.09 (0.138)

-0.03 (0.579)

-0.07 (0.222)

0.19** (0.002)

-

* significant at the 0.05 level; ** significant at the 0.01 level; Variable definitions are provided in table 4.

32

Table 6

Regression estimates of earnings quality variables on audit committee and control

variables for 309 ASX listed companies in 2001 Variable Pred.

sign

EQJones EQDD

Intercept ? 0.295

(1.902)

0.403

(3.304)**

0.286

(4.391)**

0.305

(3.925)**

ACIND - 0.006

(0.146)

-0.034

(-1.207)

0.003

(0.169)

0.014

(0.795)

ACACCEX - 0.045

(1.049)

0.031

(0.965)

-0.038

(-2.101)*

-0.040

(-1.916)*

ACLEGEX - -0.016

(-0.257)

-0.001

(-0.022)

-0.015

(-0.554)

-0.012

(-0.406)

ACMEET - -0.002

(-0.283)

-0.002

(-0.293)

0.006

(1.609)

0.003

(0.859)

ACSIZE - 0.006

(0.496)

0.005

(0.554)

-0.004

(-0.824)

-0.010

(-1.591)

AUDITOR - 0.080

(2.350)*

0.070

(2.765)**

0.001

(0.100)

0.006

(0.349)

LNTA - -0.019

(-2.140)*

-0.021

(-2.930)**

-0.012

(-3.357)**

-0.012

(-2.616)**

LEV + 0.101

(1.545)

0.015

(0.309)

0.063

(2.310)*

0.052

(1.640)

LOSS + 0.039

(1.379)

0.042

(2.025)*

0.019

(1.606)

0.012

(0.904)

LNOC + 0.023

(1.845)*

- 0.001

(0.285)

-

Capital goods ? - -0.022

(-0.379)

- -0.019

(-0.534)

Commercial, services and

supplies

? - 0.017

(0.285)

- 0.007

(0.167)

Energy ? - 0.054

(0.855)

- -0.017

(-0.419)

Food, beverage and

tobacco

? - 0.368

(6.302)**

- -0.023

(-0.629)

Healthcare equipment

and services

? - 0.012

(0.189)

- 0.004

(0.096)

Hotels, restaurants and

leisure

? - -0.014

(-0.217)

- -0.025

(-0.592)

Materials ? - 0.022

(0.410)

- -0.005

(-0.142)

Media ? - 0.242

(3.861)**

- -0.017

(-0.433)

Pharmaceuticals and

biotechnology

? - -0.037

(-0.559)

- 0.033

(0.780)

Retailing ? - -0.000

(-0.001)

- -0.008

(-0.203)

Software and services ? - 0.121

(2.029)*

- 0.031

(0.798)

Telecommunication

services

? - 0.423

(6.120)**

- 0.082

(1.861)

Adjusted R2 0.033 0.424 0.050 0.073

F statistic 1.977* 11.703** 2.469** 2.132**

N 283 306 282 305

* significant at the 0.05 level; ** significant at the 0.01 level (p-values are one-tailed when direction is as

predicted, otherwise two-tailed). Variable definitions are provided in table 4.